As you may have noticed when you logged in today, there is a new feature on the main My Fitness page. We added some back-end math to provide an estimate of future fitness. This was based on insight we obtained from running multiple Impulse-Response models on the data. The projections are using a generic version of the 3-tier model.
By looking at the training/riding you’ve been doing over the past 7 days, the system projects what your values will be in 6 weeks, based on your Peak/High/Low XSS for the week and their corresponding Training Loads.
Keep your data up-to-date for best results.
Looks great. And while you might be planning it, I think the forecasting also affected the XPMC chart.
I last rode on Tuesday, and yet my Training Load is showing slightly higher for today (Recovery Load shows holding steady). It could make sense if it was looking at the rides I’ve done in last 7 days and where I might be today (and I’m looking forward to that type of forecasting), but I should also expect the TL and RL to be dropping a bit, even with just two days off the bike.
Yes. The method uses your current load, after a drop from your previous activity and the load from a week ago, after a drop from its previous activity and uses the difference to extrapolate out 6 weeks from now. It assumes that the per system load you had over the past week, not the activities you’ve performed per se, is maintained.
But if it does not let my Recovery Load drop, then my Form will not change (much) either. At that point, any information from Form level is no longer useful, as it will probably always be quite negative, and in the ‘very tired’ level.
I get the idea of forecasting the XPMC, and look forward to seeing what I might have based on an input of expected rides (or recent rides repeated). But there also needs to be the ‘standard’ XPMC that normally devalues both Load numbers over time. Maybe one XPMC for forecasting, and another for actual readings. Or maybe a set of forecasted TL, RL, and Form numbers that could be displayed on the same XPMC chart as the standard ones.
Training Loads are the measures that align with your fitness and is what we use to project. As you point out, projecting Recovery Load is not useful out 6 weeks hence it isn’t used. Remember that we don’t really need historical load numbers to help us determine your fitness. We already know what your fitness was from your breakthroughs and best activities. Historical load numbers are useful insofar as they can help us calibrate the parameters for the prediction model such that we can get better estimates of what will happen with training and tapering that is planned.
Will we be able to load the calendar with “intended” workouts to forecast the outcome?